Hybridized particle swarm algorithm for adaptive structure training of multilayer feed-forward neural network: QSAR studies of bioactivity of organic compounds

The multilayer feed-forward ANN is an important modeling technique used in QSAR studying. The training of ANN is usually carried out only to optimize the weights of the neural network and without paying attention to the network topology. Some other strategies used to train ANN are, first, to discover an optimum structure of the network, and then to find weights for an already defined structure. These methods tend to converge to local optima, and may also lead to overfitting. In this article, a hybridized particle swarm optimization (PSO) approach was applied to the neural network structure training (HPSONN). The continuous version of PSO was used for the weight training of ANN, and the modified discrete PSO was applied to find appropriate the network architecture. The network structure and connectivity are trained simultaneously. The two versions of PSO can jointly search the global optimal ANN architecture and weights. A new objective function is formulated to determine the appropriate network architecture and optimum value of the weights. The proposed HPSONN algorithm was used to predict carcinogenic potency of aromatic amines and biological activity of a series of distamycin and distamycin-like derivatives. The results were compared to those obtained by PSO and GA training in which the network architecture was kept fixed. The comparison demonstrated that the HPSONN is a useful tool for training ANN, which converges quickly towards the optimal position, and can avoid overfitting in some extent.     

Global minimum structure searches via particle swarm optimization

Novel implementation of the evolutionary approach known as particle swarm optimization (PSO) capable of finding the global minimum of the potential energy surface of atomic assemblies is reported. This is the first time the PSO technique has been used to perform global optimization of minimum structure search for chemical systems. Significant improvements have been introduced to the original PSO algorithm to increase its efficiency and reliability and adapt it to chemical systems. The developed software has successfully found the lowest-energy structures of the LJ(26) Lennard-Jones cluster, anionic silicon hydride Si(2)H(5) (-), and triply hydrated hydroxide ion OH(-) (H(2)O)(3). It requires relatively small population sizes and demonstrates fast convergence. Efficiency of PSO has been compared with simulated annealing, and the gradient embedded genetic algorithm.     

Development of non‐bonded interaction parameters between graphene and water using particle swarm optimization

New Lennard‐Jones parameters have been developed to describe the interactions between atomistic model of graphene, represented by REBO potential, and five commonly used all‐atom water models, namely SPC, SPC/E, SPC/Fw, SPC/Fd, and TIP3P/Fs by employing particle swarm optimization (PSO) method. These new parameters were optimized to reproduce the macroscopic contact angle of water on a graphene sheet. The calculated line tension was in the order of 10⁻¹¹ J/m for the droplets of all water models. Our molecular dynamics simulations indicate the preferential orientation of water molecules near graphene–water interface with one O H bond pointing toward the graphene surface. Detailed analysis of simulation trajectories reveals the presence of water molecules with ≤∼1, ∼2, and ∼4 hydrogen bonds at the surface of air–water interface, graphene–water interface, and bulk region of the water droplet, respectively. Presence of water molecules with ≤∼1 and ∼2 hydrogen bonds suggest the existence of water clusters of different sizes at these interfaces. The trends observed in the libration, bending, and stretching bands of the vibrational spectra are closely associated with these structural features of water. The inhomogeneity in hydrogen bond network of water at the air–water and graphene–water interface is manifested by broadening of the peaks in the libration band for water present at these interfaces. The stretching band for the molecules in water droplet shows a blue shift as compared to the pure bulk water, which conjecture the presence of weaker hydrogen bond network in a droplet. © 2017 Wiley Periodicals, Inc.     

Estimation of solid vapor pressures of pure compounds at different temperatures using a multilayer network with particle swarm algorithm

Solid vapor pressures (P^S) of pure compounds have been estimated at several temperatures using a hybrid model that includes an artificial neural network with particle swarm optimization and a group contribution method. A total of 700 data points of solid vapor pressure versus temperature, corresponding to 70 substances, have been used to train the neural network developed using Matlab. The following properties were considered as input parameters: 36 structural groups, molecular mass, dipole moment, temperature and pressure in the triple point (upper limit of the sublimation curve), and the limiting value P^S → 0 as T → 0 (lower limit of the sublimation curve). Then, the solid vapor pressures of 28 other solids (280 data points) have been predicted and results compared to experimental data from the literature. The study shows that the proposed method represents an excellent alternative for the prediction of solid vapor pressures from the knowledge of some other available properties and from the structure of the molecule.     

A wavelength selection method based on random decision particle swarm optimization with attractor for near‐infrared spectral quantitative analysis

In this paper, we proposed a wavelength selection method based on random decision particle swarm optimization with attractor for near‐infrared (NIR) spectra quantitative analysis. The proposed method was incorporated with partial least square (PLS) to construct a prediction model. The proposed method chooses the current own optimal or the current global optimal to calculate the attractor. Then the particle updates its flight velocity by the attractor, and the particle state is updated by the random decision with the new velocity. Moreover, the root‐mean‐square error of cross‐validation is adopted as the fitness function for the proposed method. In order to demonstrate the usefulness of the proposed method, PLS with all wavelengths, uninformative variable elimination by PLS, elastic net, genetic algorithm combined with PLS, the discrete particle swarm optimization combined with PLS, the modified particle swarm optimization combined with PLS, the neighboring particle swarm optimization combined with PLS, and the proposed method are used for building the components quantitative analysis models of NIR spectral datasets, and the effectiveness of these models is compared. Two application studies are presented, which involve NIR data obtained from an experiment of meat content determination using NIR and a combustion procedure. Results verify that the proposed method has higher predictive ability for NIR spectral data and the number of selected wavelengths is less. The proposed method has faster convergence speed and could overcome the premature convergence problem. Furthermore, although improving the prediction precision may sacrifice the model complexity under a certain extent, the proposed method is overfitted slightly. Copyright © 2015 John Wiley & Sons, Ltd.     

A combination of modified particle swarm optimization algorithm and support vector machine for gene selection and tumor classification

In the analysis of gene expression profiles, the number of tissue samples with genes expression levels available is usually small compared with the number of genes. This can lead either to possible overfitting or even to a complete failure in analysis of microarray data. The selection of genes that are really indicative of the tissue classification concerned is becoming one of the key steps in microarray studies. In the present paper, we have combined the modified discrete particle swarm optimization (PSO) and support vector machines (SVM) for tumor classification. The modified discrete PSO is applied to select genes, while SVM is used as the classifier or the evaluator. The proposed approach is used to the microarray data of 22 normal and 40 colon tumor tissues and showed good prediction performance. It has been demonstrated that the modified PSO is a useful tool for gene selection and mining high dimension data.     

Improvement of residual bilinearization by particle swarm optimization for achieving the second‐order advantage with unfolded partial least‐squares

The combination of unfolded partial least‐squares (U‐PLS) with residual bilinearization (RBL) provides a second‐order multivariate calibration method capable of achieving the second‐order advantage. RBL is performed by varying the test sample scores in order to minimize the residues of a combined U‐PLS model for the calibrated components and a principal component model for the potential interferents. The sample scores are then employed to predict the analyte concentration, with regression coefficients taken from the calibration step. When the contribution of multiple potential interferents is severe, particle swarm optimization (PSO) helps in preventing RBL to be trapped by false minima, restoring its predictive ability and making it comparable to the standard parallel factor (PARAFAC) analysis. Both simulated and experimental systems are analyzed in order to show the potentiality of the new technique. Copyright © 2007 John Wiley & Sons, Ltd.     

Estimation of rotation ambiguity in multivariate curve resolution with charged particle swarm optimization (cPSO‐MCR)

Rotation ambiguity (RA) in multivariate curve resolution (MCR) is an undesirable case, when the physicochemical constraints are not sufficiently strong to provide a unique resolution of the data matrix of the mixtures into spectra and concentration profiles of individual chemical components. RA is often met in MCR of overlapped chromatographic peaks, kinetic and equilibrium data, and fluorescence two‐dimensional spectra. In case of RA, a single candidate solution has little practical value. So, the whole set of feasible solutions should be characterized somehow. It is a quite intricate task in a general case. In the present paper, a method was proposed to estimate RA with charged particle swarm optimization (cPSO), a population‐based algorithm. The criteria for updating the particles were modified, so that the swarm converged to the steady state, which spanned the set of feasible solutions. The performance of cPSO‐MCR was demonstrated on test functions, simulated datasets, and real‐world data. Good accordance of the cPSO‐MCR results with the analytical solutions (Borgen plots) was observed. cPSO‐MCR was also shown to be capable of estimating the strength of the constraints and of revealing RA in noisy data. As compared with analytical methods, cPSO‐MCR is simpler to implement, expands to more than three chemical compounds, is immune to noise, and can be easily adapted to virtually all types of constraints and objective functions (constraint based or residue based). cPSO‐MCR also provides natural visual information about the level of RA in spectra and concentration profiles, similar to the methods of two extreme solutions (e.g., MCR‐BANDS). Copyright © 2014 John Wiley & Sons, Ltd.     

Generalized Gaussian reference curve measurement model for high‐performance liquid chromatography with diode array detector separation and its solution by multi‐target intermittent particle swarm optimization

In order to separate a high‐performance liquid chromatography with diode array detector (HPLC‐DAD) data set to chromatogram peaks and spectra for all compounds, a separation method based on the model of generalized Gaussian reference curve measurement (GGRCM) and the algorithm of multi‐target intermittent particle swarm optimization (MIPSO) is proposed in this paper. A parameter θ is constructed to generate a reference curve r(θ) for a chromatogram peak based on its physical principle. The GGRCM model is proposed to calculate the fitness ε(θ) for every θ, which indicates the possibility for the HPLC‐DAD data set to contain a chromatogram peak similar to the r(θ). The smaller the fitness is, the higher the possibility. The algorithm of MIPSO is then introduced to calculate the optimal parameters by minimizing the fitness mentioned earlier. Finally, chromatogram peaks are constructed based on these optimal parameters, and the spectra are calculated by an estimator. Through the simulations and experiments, the following conclusions are drawn: (i) the GGRCM‐MIPSO method can extract chromatogram peaks from simulation data set without knowing the number of the compounds in advance even when a severe overlap and white noise exist and (ii) the GGRCM‐MIPSO method can be applied to the real HPLC‐DAD data set. Copyright © 2014 John Wiley & Sons, Ltd.     

Partial least‐squares discriminant analysis optimized by particle swarm optimization: application to 1H nuclear magnetic resonance analysis of lung cancer metabonomics

The complexity of metabolic profiles makes chemometric tools indispensable for extracting the most significant information. Partial least‐squares discriminant analysis (PLS‐DA) acts as one of the most effective strategies for data analysis in metabonomics. However, its actual efficacy in metabonomics is often weakened by the high similarity of metabolic profiles, which contain excessive variables. To rectify this situation, particle swarm optimization (PSO) was introduced to improve PLS‐DA by simultaneously selecting the optimal sample and variable subsets, the appropriate variable weights, and the best number of latent variables (SVWL) in PLS‐DA, forming a new algorithm named PSO‐SVWL‐PLSDA. Combined with ¹H nuclear magnetic resonance‐based metabonomics, PSO‐SVWL‐PLSDA was applied to recognize the patients with lung cancer from the healthy controls. PLS‐DA was also investigated as a comparison. Relatively to the recognition rates of 86% and 65%, which were yielded by PLS‐DA, respectively, for the training and test sets, those of 98.3% and 90% were offered by PSO‐SVWL‐PLSDA. Moreover, several most discriminative metabolites were identified by PSO‐SVWL‐PLSDA to aid the diagnosis of lung cancer, including lactate, glucose (α‐glucose and β‐glucose), threonine, valine, taurine, trimethylamine, glutamine, glycoprotein, proline, and lipid. Copyright © 2015 John Wiley & Sons, Ltd.     

Hybrid particle swarm optimization and tabu search approach for selecting genes for tumor classification using gene expression data

Gene expression data are characterized by thousands even tens of thousands of measured genes on only a few tissue samples. This can lead either to possible overfitting and dimensional curse or even to a complete failure in analysis of microarray data. Gene selection is an important component for gene expression-based tumor classification systems. In this paper, we develop a hybrid particle swarm optimization (PSO) and tabu search (HPSOTS) approach for gene selection for tumor classification. The incorporation of tabu search (TS) as a local improvement procedure enables the algorithm HPSOTS to overleap local optima and show satisfactory performance. The proposed approach is applied to three different microarray data sets. Moreover, we compare the performance of HPSOTS on these datasets to that of stepwise selection, the pure TS and PSO algorithm. It has been demonstrated that the HPSOTS is a useful tool for gene selection and mining high dimension data.     

Evaluation of repulsive particle swarm method for phase equilibrium and phase stability problems

Phase equilibrium and stability problems are of crucial importance in simulation, design and optimization of several separation processes. Recently, these problems have been solved using minimization of Gibbs free energy, using global optimization techniques. In this paper, repulsive particle swarm (RPS), a recent global optimization technique is explored for the solution of phase stability and phase equilibrium.     

Optimization of bimetallic Cu–Au and Ag–Au clusters by using a modified adaptive immune optimization algorithm

A modified adaptive immune optimization algorithm (AIOA) is designed for optimization of Cu–Au and Ag–Au bimetallic clusters with Gupta potential. Compared with homoatom clusters, there are homotopic isomers in bimetallic cluster, so atom exchange operation is presented in the modified AIOA. The efficiency of the algorithm is tested by optimization of Cu_nAu_38‐n (0 ≤ n ≤ 38). Results show that all the structures with the putative global minimal energies are successfully located. In the optimization of Ag_nAu_55‐n (0 ≤ n ≤ 55) bimetallic clusters, all the structures with the reported minimal energies are obtained, and 36 structures with even lower potential energies are found. On the other hand, with the optimized structures of Cu_nAu_55‐n, it is shown that all 55‐atom Cu–Au bimetallic clusters are Mackay icosahedra except for Au₅₅, which is a face‐centered cubic (fcc)‐like structure; Cu₅₅, Cu₁₂Au₄₃, and Cu₁Au₅₄ have two‐shell Mackay icosahedral geometries with I_h point group symmetry. © 2009 Wiley Periodicals, Inc. J Comput Chem 2009     

A comparative study of particle swarm optimization and its variants for phase stability and equilibrium calculations in multicomponent reactive and non-reactive systems

Particle swarm optimization is a novel evolutionary stochastic global optimization method that has gained popularity in the chemical engineering community. This optimization strategy has been successfully used for several applications including thermodynamic calculations. To the best of our knowledge, the performance of PSO in phase stability and equilibrium calculations for both multicomponent reactive and non-reactive mixtures has not yet been reported. This study introduces the application of particle swarm optimization and several of its variants for solving phase stability and equilibrium problems in multicomponent systems with or without chemical equilibrium. The reliability and efficiency of a number of particle swarm optimization algorithms are tested and compared using multicomponent systems with vapor–liquid and liquid–liquid equilibrium. Our results indicate that the classical particle swarm optimization with constant cognitive and social parameters is a reliable method and offers the best performance for global minimization of the tangent plane distance function and the Gibbs energy function in both reactive and non-reactive systems.     

基于混沌自适应粒子群人工神经网络的气体在聚合物中的溶解模型

为提高溶解预测模型的效率和关联度, 建立基于混沌理论、自适应粒子群优化(PSO)算法和反向传播(BP)算法的混沌自适应PSO-BP神经网络模型, 并对二氧化碳(CO₂)在聚苯乙烯(PS)和聚丙烯(PP)中、氮气(N₂)在PS中的溶解度进行预测试验. 模型选用压力和温度作为输入参数, 使用试探法确定隐含层结点个数为8, 输出为预测的溶解度. 模型融合混沌理论、自适应PSO和BP算法各自的优势, 提高了训练速度和预测精度. 结果表明, 混沌自适应PSO-BP神经网络有很好的预测能力, 预测值与实验值相当吻合, 通过与传统BP神经网络和PSO-BP神经网络的比较可知, 其预测精度和相关性均明显较优, 预测平均绝对误差(AAD), 标准偏差(SD)和平方相关系数(R²)分别为0.0058, 0.0198和0.9914.     

Qualitative and quantitative analysis of the chemical constituents in Mahuang‐Fuzi‐Xixin decoction based on high performance liquid chromatography combined with time‐of‐flight mass spectrometry and triple quadrupole mass spectrometers

High‐performance liquid chromatography coupled with time‐of‐flight mass spectrometry (HPLC‐TOF/MS) and high‐performance liquid chromatography–triple quadrupole mass spectrometry (HPLC‐QQQ/MS/MS) were utilized to clarify the chemical constituents of Mahuang‐Fuzi‐Xixin Decoction. There are 52 compounds, including alkaloids, amino acids and organic acids were identified or tentatively characterized by their characteristic high resolution mass data by HPLC‐QQQ/MS/MS. In the subsequent quantitative analysis, 10 constituents, including methyl ephedrine, aconine, songrine, fuziline, neoline, talatisamine, chasmanine, benzoylmesaconine, benzoylaconine and benzoylhypaconine were simultaneously determined by HPLC‐QQQ/MS/MS with multiple reaction monitoring mode. Satisfactory linearity was achieved with wide linear range and fine determination coefficient (r > 0.9992). The relative standard deviations (RSD) of inter‐ and intra‐day precisions were <3%. This method was also validated by repeatability, stability and recovery with RSD <3% respectively. A highly sensitive and efficient method was established for chemical constituents studying, including identification and quantification of Mahuang‐Fuzi‐Xixin decoction.